In this paper, a new algorithm to recognize characters in vehicle license plate based on hidden Markov model(HMM) is presented. In the method, two-dimensional hidden Markov model(2D-HMM) is used to recognize chinese characters, and pseudo two-dimensional hidden Markov model(P2D-HMM) is used to recognize English characters and numbers. It is proved that this algorithm can be used to recognize varisized characters with high tolerance to distortion and blur, and its character recognition accuracy is above 94%. It can be put into practical use.